sam1120 commited on
Commit
17321c5
Β·
1 Parent(s): 0d1f585

Training in progress, step 240

Browse files
{checkpoint-40 β†’ checkpoint-240}/config.json RENAMED
File without changes
{checkpoint-40 β†’ checkpoint-240}/optimizer.pt RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c571df6c9d787b508470756f114a40f34a3fbad0666afd2f30bad1f9b7e9b322
3
  size 677455103
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eeb7af946a1afe6044dc7aed9b3473f98f02cba027fe9f4be1b46b2febe3d654
3
  size 677455103
{checkpoint-40 β†’ checkpoint-240}/pytorch_model.bin RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8305ac856f53ce31a7bd13e9a255e9677351e9638293e50461bc81d8866ecf5d
3
  size 338796489
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7123f1ef46597d6e8e27beb20eeb550ccffbb057b64c28f5346b0a7cf308bc60
3
  size 338796489
{checkpoint-40 β†’ checkpoint-240}/rng_state.pth RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d781c688df7712c332ad714f724024c0acaa00e2fb96df0253fd63a2abb02e48
3
  size 14575
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e928c4745ae7c69f815d085a954590c856da58b062aa9b31e7a3fb202f8e9ba5
3
  size 14575
{checkpoint-40 β†’ checkpoint-240}/scheduler.pt RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:1c0ed14feec3de08a3859fe296b28c00d6193830a814756c3be0ca28a2c9ef2d
3
  size 627
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ee62c4ee2fb977809bf14c25643968ad4911b988e326c2d39b9eb3ce2a0defe4
3
  size 627
checkpoint-240/trainer_state.json ADDED
@@ -0,0 +1,1864 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.3769916594028473,
3
+ "best_model_checkpoint": "/robodata/smodak/Projects/nspl/scripts/terrainseg/training/models/dropoff-utcustom-train-SF-RGB-b5_3/checkpoint-240",
4
+ "epoch": 120.0,
5
+ "global_step": 240,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.5,
12
+ "learning_rate": 4.1666666666666667e-07,
13
+ "loss": 1.3484,
14
+ "step": 1
15
+ },
16
+ {
17
+ "epoch": 1.0,
18
+ "learning_rate": 8.333333333333333e-07,
19
+ "loss": 1.3481,
20
+ "step": 2
21
+ },
22
+ {
23
+ "epoch": 1.5,
24
+ "learning_rate": 1.25e-06,
25
+ "loss": 1.3531,
26
+ "step": 3
27
+ },
28
+ {
29
+ "epoch": 2.0,
30
+ "learning_rate": 1.6666666666666667e-06,
31
+ "loss": 1.3399,
32
+ "step": 4
33
+ },
34
+ {
35
+ "epoch": 2.5,
36
+ "learning_rate": 2.0833333333333334e-06,
37
+ "loss": 1.3398,
38
+ "step": 5
39
+ },
40
+ {
41
+ "epoch": 3.0,
42
+ "learning_rate": 2.5e-06,
43
+ "loss": 1.3386,
44
+ "step": 6
45
+ },
46
+ {
47
+ "epoch": 3.5,
48
+ "learning_rate": 2.916666666666667e-06,
49
+ "loss": 1.3376,
50
+ "step": 7
51
+ },
52
+ {
53
+ "epoch": 4.0,
54
+ "learning_rate": 3.3333333333333333e-06,
55
+ "loss": 1.3249,
56
+ "step": 8
57
+ },
58
+ {
59
+ "epoch": 4.5,
60
+ "learning_rate": 3.7500000000000005e-06,
61
+ "loss": 1.3205,
62
+ "step": 9
63
+ },
64
+ {
65
+ "epoch": 5.0,
66
+ "learning_rate": 4.166666666666667e-06,
67
+ "loss": 1.3135,
68
+ "step": 10
69
+ },
70
+ {
71
+ "epoch": 5.0,
72
+ "eval_accuracy_dropoff": 0.4068717172412211,
73
+ "eval_accuracy_undropoff": 0.11034322178517518,
74
+ "eval_accuracy_unlabeled": NaN,
75
+ "eval_iou_dropoff": 0.053473330857492875,
76
+ "eval_iou_undropoff": 0.1102453041626867,
77
+ "eval_iou_unlabeled": 0.0,
78
+ "eval_loss": 1.2007640600204468,
79
+ "eval_mean_accuracy": 0.25860746951319813,
80
+ "eval_mean_iou": 0.05457287834005986,
81
+ "eval_overall_accuracy": 0.12267265319824219,
82
+ "eval_runtime": 2.5511,
83
+ "eval_samples_per_second": 7.84,
84
+ "eval_steps_per_second": 0.784,
85
+ "step": 10
86
+ },
87
+ {
88
+ "epoch": 5.5,
89
+ "learning_rate": 4.583333333333333e-06,
90
+ "loss": 1.3089,
91
+ "step": 11
92
+ },
93
+ {
94
+ "epoch": 6.0,
95
+ "learning_rate": 5e-06,
96
+ "loss": 1.3043,
97
+ "step": 12
98
+ },
99
+ {
100
+ "epoch": 6.5,
101
+ "learning_rate": 4.978070175438597e-06,
102
+ "loss": 1.2871,
103
+ "step": 13
104
+ },
105
+ {
106
+ "epoch": 7.0,
107
+ "learning_rate": 4.956140350877193e-06,
108
+ "loss": 1.2832,
109
+ "step": 14
110
+ },
111
+ {
112
+ "epoch": 7.5,
113
+ "learning_rate": 4.9342105263157895e-06,
114
+ "loss": 1.2709,
115
+ "step": 15
116
+ },
117
+ {
118
+ "epoch": 8.0,
119
+ "learning_rate": 4.912280701754386e-06,
120
+ "loss": 1.2634,
121
+ "step": 16
122
+ },
123
+ {
124
+ "epoch": 8.5,
125
+ "learning_rate": 4.890350877192983e-06,
126
+ "loss": 1.257,
127
+ "step": 17
128
+ },
129
+ {
130
+ "epoch": 9.0,
131
+ "learning_rate": 4.8684210526315795e-06,
132
+ "loss": 1.2403,
133
+ "step": 18
134
+ },
135
+ {
136
+ "epoch": 9.5,
137
+ "learning_rate": 4.846491228070176e-06,
138
+ "loss": 1.2398,
139
+ "step": 19
140
+ },
141
+ {
142
+ "epoch": 10.0,
143
+ "learning_rate": 4.824561403508772e-06,
144
+ "loss": 1.2309,
145
+ "step": 20
146
+ },
147
+ {
148
+ "epoch": 10.0,
149
+ "eval_accuracy_dropoff": 0.43877153145714354,
150
+ "eval_accuracy_undropoff": 0.24072332003617994,
151
+ "eval_accuracy_unlabeled": NaN,
152
+ "eval_iou_dropoff": 0.11288106949650205,
153
+ "eval_iou_undropoff": 0.24004849364289937,
154
+ "eval_iou_unlabeled": 0.0,
155
+ "eval_loss": 1.1293678283691406,
156
+ "eval_mean_accuracy": 0.3397474257466617,
157
+ "eval_mean_iou": 0.11764318771313381,
158
+ "eval_overall_accuracy": 0.24895801544189453,
159
+ "eval_runtime": 2.6298,
160
+ "eval_samples_per_second": 7.605,
161
+ "eval_steps_per_second": 0.761,
162
+ "step": 20
163
+ },
164
+ {
165
+ "epoch": 10.5,
166
+ "learning_rate": 4.802631578947369e-06,
167
+ "loss": 1.2159,
168
+ "step": 21
169
+ },
170
+ {
171
+ "epoch": 11.0,
172
+ "learning_rate": 4.780701754385965e-06,
173
+ "loss": 1.2243,
174
+ "step": 22
175
+ },
176
+ {
177
+ "epoch": 11.5,
178
+ "learning_rate": 4.758771929824561e-06,
179
+ "loss": 1.2094,
180
+ "step": 23
181
+ },
182
+ {
183
+ "epoch": 12.0,
184
+ "learning_rate": 4.736842105263158e-06,
185
+ "loss": 1.1904,
186
+ "step": 24
187
+ },
188
+ {
189
+ "epoch": 12.5,
190
+ "learning_rate": 4.714912280701755e-06,
191
+ "loss": 1.1865,
192
+ "step": 25
193
+ },
194
+ {
195
+ "epoch": 13.0,
196
+ "learning_rate": 4.692982456140351e-06,
197
+ "loss": 1.1715,
198
+ "step": 26
199
+ },
200
+ {
201
+ "epoch": 13.5,
202
+ "learning_rate": 4.671052631578948e-06,
203
+ "loss": 1.1722,
204
+ "step": 27
205
+ },
206
+ {
207
+ "epoch": 14.0,
208
+ "learning_rate": 4.649122807017544e-06,
209
+ "loss": 1.1511,
210
+ "step": 28
211
+ },
212
+ {
213
+ "epoch": 14.5,
214
+ "learning_rate": 4.6271929824561405e-06,
215
+ "loss": 1.1512,
216
+ "step": 29
217
+ },
218
+ {
219
+ "epoch": 15.0,
220
+ "learning_rate": 4.605263157894737e-06,
221
+ "loss": 1.1346,
222
+ "step": 30
223
+ },
224
+ {
225
+ "epoch": 15.0,
226
+ "eval_accuracy_dropoff": 0.46940067432739285,
227
+ "eval_accuracy_undropoff": 0.5035900324087019,
228
+ "eval_accuracy_unlabeled": NaN,
229
+ "eval_iou_dropoff": 0.15243042262584855,
230
+ "eval_iou_undropoff": 0.4989255427443525,
231
+ "eval_iou_unlabeled": 0.0,
232
+ "eval_loss": 1.039512276649475,
233
+ "eval_mean_accuracy": 0.4864953533680474,
234
+ "eval_mean_iou": 0.21711865512340034,
235
+ "eval_overall_accuracy": 0.5021684646606446,
236
+ "eval_runtime": 2.9463,
237
+ "eval_samples_per_second": 6.788,
238
+ "eval_steps_per_second": 0.679,
239
+ "step": 30
240
+ },
241
+ {
242
+ "epoch": 15.5,
243
+ "learning_rate": 4.583333333333333e-06,
244
+ "loss": 1.1343,
245
+ "step": 31
246
+ },
247
+ {
248
+ "epoch": 16.0,
249
+ "learning_rate": 4.56140350877193e-06,
250
+ "loss": 1.1182,
251
+ "step": 32
252
+ },
253
+ {
254
+ "epoch": 16.5,
255
+ "learning_rate": 4.539473684210527e-06,
256
+ "loss": 1.132,
257
+ "step": 33
258
+ },
259
+ {
260
+ "epoch": 17.0,
261
+ "learning_rate": 4.517543859649123e-06,
262
+ "loss": 1.0986,
263
+ "step": 34
264
+ },
265
+ {
266
+ "epoch": 17.5,
267
+ "learning_rate": 4.4956140350877196e-06,
268
+ "loss": 1.0887,
269
+ "step": 35
270
+ },
271
+ {
272
+ "epoch": 18.0,
273
+ "learning_rate": 4.473684210526316e-06,
274
+ "loss": 1.1214,
275
+ "step": 36
276
+ },
277
+ {
278
+ "epoch": 18.5,
279
+ "learning_rate": 4.451754385964912e-06,
280
+ "loss": 1.0945,
281
+ "step": 37
282
+ },
283
+ {
284
+ "epoch": 19.0,
285
+ "learning_rate": 4.429824561403509e-06,
286
+ "loss": 1.076,
287
+ "step": 38
288
+ },
289
+ {
290
+ "epoch": 19.5,
291
+ "learning_rate": 4.407894736842105e-06,
292
+ "loss": 1.0444,
293
+ "step": 39
294
+ },
295
+ {
296
+ "epoch": 20.0,
297
+ "learning_rate": 4.385964912280702e-06,
298
+ "loss": 1.1088,
299
+ "step": 40
300
+ },
301
+ {
302
+ "epoch": 20.0,
303
+ "eval_accuracy_dropoff": 0.48076332025963897,
304
+ "eval_accuracy_undropoff": 0.6235115828521449,
305
+ "eval_accuracy_unlabeled": NaN,
306
+ "eval_iou_dropoff": 0.1660792839847365,
307
+ "eval_iou_undropoff": 0.616259059512146,
308
+ "eval_iou_unlabeled": 0.0,
309
+ "eval_loss": 0.9755474328994751,
310
+ "eval_mean_accuracy": 0.5521374515558919,
311
+ "eval_mean_iou": 0.2607794478322942,
312
+ "eval_overall_accuracy": 0.6175762176513672,
313
+ "eval_runtime": 3.0937,
314
+ "eval_samples_per_second": 6.465,
315
+ "eval_steps_per_second": 0.646,
316
+ "step": 40
317
+ },
318
+ {
319
+ "epoch": 20.5,
320
+ "learning_rate": 4.364035087719299e-06,
321
+ "loss": 1.0507,
322
+ "step": 41
323
+ },
324
+ {
325
+ "epoch": 21.0,
326
+ "learning_rate": 4.342105263157895e-06,
327
+ "loss": 1.0644,
328
+ "step": 42
329
+ },
330
+ {
331
+ "epoch": 21.5,
332
+ "learning_rate": 4.320175438596491e-06,
333
+ "loss": 1.0289,
334
+ "step": 43
335
+ },
336
+ {
337
+ "epoch": 22.0,
338
+ "learning_rate": 4.298245614035088e-06,
339
+ "loss": 1.0551,
340
+ "step": 44
341
+ },
342
+ {
343
+ "epoch": 22.5,
344
+ "learning_rate": 4.276315789473684e-06,
345
+ "loss": 1.0025,
346
+ "step": 45
347
+ },
348
+ {
349
+ "epoch": 23.0,
350
+ "learning_rate": 4.254385964912281e-06,
351
+ "loss": 1.0486,
352
+ "step": 46
353
+ },
354
+ {
355
+ "epoch": 23.5,
356
+ "learning_rate": 4.232456140350878e-06,
357
+ "loss": 1.019,
358
+ "step": 47
359
+ },
360
+ {
361
+ "epoch": 24.0,
362
+ "learning_rate": 4.210526315789474e-06,
363
+ "loss": 0.9953,
364
+ "step": 48
365
+ },
366
+ {
367
+ "epoch": 24.5,
368
+ "learning_rate": 4.1885964912280705e-06,
369
+ "loss": 0.9797,
370
+ "step": 49
371
+ },
372
+ {
373
+ "epoch": 25.0,
374
+ "learning_rate": 4.166666666666667e-06,
375
+ "loss": 1.007,
376
+ "step": 50
377
+ },
378
+ {
379
+ "epoch": 25.0,
380
+ "eval_accuracy_dropoff": 0.506777678387119,
381
+ "eval_accuracy_undropoff": 0.6849468196784603,
382
+ "eval_accuracy_unlabeled": NaN,
383
+ "eval_iou_dropoff": 0.19228861306779316,
384
+ "eval_iou_undropoff": 0.6762847927344823,
385
+ "eval_iou_unlabeled": 0.0,
386
+ "eval_loss": 0.9196955561637878,
387
+ "eval_mean_accuracy": 0.5958622490327896,
388
+ "eval_mean_iou": 0.2895244686007585,
389
+ "eval_overall_accuracy": 0.6775386810302735,
390
+ "eval_runtime": 2.8681,
391
+ "eval_samples_per_second": 6.973,
392
+ "eval_steps_per_second": 0.697,
393
+ "step": 50
394
+ },
395
+ {
396
+ "epoch": 25.5,
397
+ "learning_rate": 4.144736842105263e-06,
398
+ "loss": 0.9886,
399
+ "step": 51
400
+ },
401
+ {
402
+ "epoch": 26.0,
403
+ "learning_rate": 4.12280701754386e-06,
404
+ "loss": 0.9687,
405
+ "step": 52
406
+ },
407
+ {
408
+ "epoch": 26.5,
409
+ "learning_rate": 4.100877192982457e-06,
410
+ "loss": 0.9731,
411
+ "step": 53
412
+ },
413
+ {
414
+ "epoch": 27.0,
415
+ "learning_rate": 4.078947368421053e-06,
416
+ "loss": 0.9503,
417
+ "step": 54
418
+ },
419
+ {
420
+ "epoch": 27.5,
421
+ "learning_rate": 4.05701754385965e-06,
422
+ "loss": 0.9563,
423
+ "step": 55
424
+ },
425
+ {
426
+ "epoch": 28.0,
427
+ "learning_rate": 4.035087719298246e-06,
428
+ "loss": 0.9525,
429
+ "step": 56
430
+ },
431
+ {
432
+ "epoch": 28.5,
433
+ "learning_rate": 4.013157894736842e-06,
434
+ "loss": 0.911,
435
+ "step": 57
436
+ },
437
+ {
438
+ "epoch": 29.0,
439
+ "learning_rate": 3.991228070175439e-06,
440
+ "loss": 0.9646,
441
+ "step": 58
442
+ },
443
+ {
444
+ "epoch": 29.5,
445
+ "learning_rate": 3.969298245614036e-06,
446
+ "loss": 0.9424,
447
+ "step": 59
448
+ },
449
+ {
450
+ "epoch": 30.0,
451
+ "learning_rate": 3.947368421052632e-06,
452
+ "loss": 0.9145,
453
+ "step": 60
454
+ },
455
+ {
456
+ "epoch": 30.0,
457
+ "eval_accuracy_dropoff": 0.5167962567948806,
458
+ "eval_accuracy_undropoff": 0.7429246639475331,
459
+ "eval_accuracy_unlabeled": NaN,
460
+ "eval_iou_dropoff": 0.21555466691667607,
461
+ "eval_iou_undropoff": 0.7329009745349515,
462
+ "eval_iou_unlabeled": 0.0,
463
+ "eval_loss": 0.8634790182113647,
464
+ "eval_mean_accuracy": 0.6298604603712068,
465
+ "eval_mean_iou": 0.3161518804838759,
466
+ "eval_overall_accuracy": 0.7335224151611328,
467
+ "eval_runtime": 2.85,
468
+ "eval_samples_per_second": 7.018,
469
+ "eval_steps_per_second": 0.702,
470
+ "step": 60
471
+ },
472
+ {
473
+ "epoch": 30.5,
474
+ "learning_rate": 3.925438596491229e-06,
475
+ "loss": 0.9084,
476
+ "step": 61
477
+ },
478
+ {
479
+ "epoch": 31.0,
480
+ "learning_rate": 3.903508771929825e-06,
481
+ "loss": 0.9234,
482
+ "step": 62
483
+ },
484
+ {
485
+ "epoch": 31.5,
486
+ "learning_rate": 3.8815789473684214e-06,
487
+ "loss": 0.9259,
488
+ "step": 63
489
+ },
490
+ {
491
+ "epoch": 32.0,
492
+ "learning_rate": 3.859649122807018e-06,
493
+ "loss": 0.8798,
494
+ "step": 64
495
+ },
496
+ {
497
+ "epoch": 32.5,
498
+ "learning_rate": 3.837719298245615e-06,
499
+ "loss": 0.8789,
500
+ "step": 65
501
+ },
502
+ {
503
+ "epoch": 33.0,
504
+ "learning_rate": 3.815789473684211e-06,
505
+ "loss": 0.8941,
506
+ "step": 66
507
+ },
508
+ {
509
+ "epoch": 33.5,
510
+ "learning_rate": 3.7938596491228073e-06,
511
+ "loss": 0.8674,
512
+ "step": 67
513
+ },
514
+ {
515
+ "epoch": 34.0,
516
+ "learning_rate": 3.7719298245614037e-06,
517
+ "loss": 0.8831,
518
+ "step": 68
519
+ },
520
+ {
521
+ "epoch": 34.5,
522
+ "learning_rate": 3.7500000000000005e-06,
523
+ "loss": 0.8512,
524
+ "step": 69
525
+ },
526
+ {
527
+ "epoch": 35.0,
528
+ "learning_rate": 3.728070175438597e-06,
529
+ "loss": 0.8745,
530
+ "step": 70
531
+ },
532
+ {
533
+ "epoch": 35.0,
534
+ "eval_accuracy_dropoff": 0.5666827220807816,
535
+ "eval_accuracy_undropoff": 0.7900781808936921,
536
+ "eval_accuracy_unlabeled": NaN,
537
+ "eval_iou_dropoff": 0.2404217803052419,
538
+ "eval_iou_undropoff": 0.7790981497740644,
539
+ "eval_iou_unlabeled": 0.0,
540
+ "eval_loss": 0.8070214986801147,
541
+ "eval_mean_accuracy": 0.6783804514872369,
542
+ "eval_mean_iou": 0.3398399766931021,
543
+ "eval_overall_accuracy": 0.780789566040039,
544
+ "eval_runtime": 2.9903,
545
+ "eval_samples_per_second": 6.688,
546
+ "eval_steps_per_second": 0.669,
547
+ "step": 70
548
+ },
549
+ {
550
+ "epoch": 35.5,
551
+ "learning_rate": 3.7061403508771933e-06,
552
+ "loss": 0.8354,
553
+ "step": 71
554
+ },
555
+ {
556
+ "epoch": 36.0,
557
+ "learning_rate": 3.6842105263157896e-06,
558
+ "loss": 0.8436,
559
+ "step": 72
560
+ },
561
+ {
562
+ "epoch": 36.5,
563
+ "learning_rate": 3.662280701754386e-06,
564
+ "loss": 0.8022,
565
+ "step": 73
566
+ },
567
+ {
568
+ "epoch": 37.0,
569
+ "learning_rate": 3.640350877192983e-06,
570
+ "loss": 0.8872,
571
+ "step": 74
572
+ },
573
+ {
574
+ "epoch": 37.5,
575
+ "learning_rate": 3.618421052631579e-06,
576
+ "loss": 0.8419,
577
+ "step": 75
578
+ },
579
+ {
580
+ "epoch": 38.0,
581
+ "learning_rate": 3.596491228070176e-06,
582
+ "loss": 0.8014,
583
+ "step": 76
584
+ },
585
+ {
586
+ "epoch": 38.5,
587
+ "learning_rate": 3.5745614035087724e-06,
588
+ "loss": 0.8142,
589
+ "step": 77
590
+ },
591
+ {
592
+ "epoch": 39.0,
593
+ "learning_rate": 3.5526315789473687e-06,
594
+ "loss": 0.7952,
595
+ "step": 78
596
+ },
597
+ {
598
+ "epoch": 39.5,
599
+ "learning_rate": 3.530701754385965e-06,
600
+ "loss": 0.7875,
601
+ "step": 79
602
+ },
603
+ {
604
+ "epoch": 40.0,
605
+ "learning_rate": 3.5087719298245615e-06,
606
+ "loss": 0.8088,
607
+ "step": 80
608
+ },
609
+ {
610
+ "epoch": 40.0,
611
+ "eval_accuracy_dropoff": 0.5992797999954127,
612
+ "eval_accuracy_undropoff": 0.8389431797941644,
613
+ "eval_accuracy_unlabeled": NaN,
614
+ "eval_iou_dropoff": 0.27303754266211605,
615
+ "eval_iou_undropoff": 0.8271554915987344,
616
+ "eval_iou_unlabeled": 0.0,
617
+ "eval_loss": 0.7441512942314148,
618
+ "eval_mean_accuracy": 0.7191114898947886,
619
+ "eval_mean_iou": 0.36673101142028347,
620
+ "eval_overall_accuracy": 0.828978157043457,
621
+ "eval_runtime": 2.8979,
622
+ "eval_samples_per_second": 6.902,
623
+ "eval_steps_per_second": 0.69,
624
+ "step": 80
625
+ },
626
+ {
627
+ "epoch": 40.5,
628
+ "learning_rate": 3.486842105263158e-06,
629
+ "loss": 0.7679,
630
+ "step": 81
631
+ },
632
+ {
633
+ "epoch": 41.0,
634
+ "learning_rate": 3.464912280701755e-06,
635
+ "loss": 0.8128,
636
+ "step": 82
637
+ },
638
+ {
639
+ "epoch": 41.5,
640
+ "learning_rate": 3.4429824561403515e-06,
641
+ "loss": 0.805,
642
+ "step": 83
643
+ },
644
+ {
645
+ "epoch": 42.0,
646
+ "learning_rate": 3.421052631578948e-06,
647
+ "loss": 0.7599,
648
+ "step": 84
649
+ },
650
+ {
651
+ "epoch": 42.5,
652
+ "learning_rate": 3.399122807017544e-06,
653
+ "loss": 0.7712,
654
+ "step": 85
655
+ },
656
+ {
657
+ "epoch": 43.0,
658
+ "learning_rate": 3.3771929824561406e-06,
659
+ "loss": 0.7746,
660
+ "step": 86
661
+ },
662
+ {
663
+ "epoch": 43.5,
664
+ "learning_rate": 3.355263157894737e-06,
665
+ "loss": 0.7474,
666
+ "step": 87
667
+ },
668
+ {
669
+ "epoch": 44.0,
670
+ "learning_rate": 3.3333333333333333e-06,
671
+ "loss": 0.7928,
672
+ "step": 88
673
+ },
674
+ {
675
+ "epoch": 44.5,
676
+ "learning_rate": 3.3114035087719297e-06,
677
+ "loss": 0.7545,
678
+ "step": 89
679
+ },
680
+ {
681
+ "epoch": 45.0,
682
+ "learning_rate": 3.289473684210527e-06,
683
+ "loss": 0.7184,
684
+ "step": 90
685
+ },
686
+ {
687
+ "epoch": 45.0,
688
+ "eval_accuracy_dropoff": 0.6322713823711553,
689
+ "eval_accuracy_undropoff": 0.8702378263383143,
690
+ "eval_accuracy_unlabeled": NaN,
691
+ "eval_iou_dropoff": 0.2914895506858354,
692
+ "eval_iou_undropoff": 0.8580097549336819,
693
+ "eval_iou_unlabeled": 0.0,
694
+ "eval_loss": 0.6956223249435425,
695
+ "eval_mean_accuracy": 0.7512546043547348,
696
+ "eval_mean_iou": 0.3831664352065058,
697
+ "eval_overall_accuracy": 0.8603433609008789,
698
+ "eval_runtime": 2.9361,
699
+ "eval_samples_per_second": 6.812,
700
+ "eval_steps_per_second": 0.681,
701
+ "step": 90
702
+ },
703
+ {
704
+ "epoch": 45.5,
705
+ "learning_rate": 3.2675438596491233e-06,
706
+ "loss": 0.7254,
707
+ "step": 91
708
+ },
709
+ {
710
+ "epoch": 46.0,
711
+ "learning_rate": 3.2456140350877197e-06,
712
+ "loss": 0.7387,
713
+ "step": 92
714
+ },
715
+ {
716
+ "epoch": 46.5,
717
+ "learning_rate": 3.223684210526316e-06,
718
+ "loss": 0.7656,
719
+ "step": 93
720
+ },
721
+ {
722
+ "epoch": 47.0,
723
+ "learning_rate": 3.2017543859649124e-06,
724
+ "loss": 0.6769,
725
+ "step": 94
726
+ },
727
+ {
728
+ "epoch": 47.5,
729
+ "learning_rate": 3.179824561403509e-06,
730
+ "loss": 0.7103,
731
+ "step": 95
732
+ },
733
+ {
734
+ "epoch": 48.0,
735
+ "learning_rate": 3.157894736842105e-06,
736
+ "loss": 0.7061,
737
+ "step": 96
738
+ },
739
+ {
740
+ "epoch": 48.5,
741
+ "learning_rate": 3.135964912280702e-06,
742
+ "loss": 0.6942,
743
+ "step": 97
744
+ },
745
+ {
746
+ "epoch": 49.0,
747
+ "learning_rate": 3.1140350877192988e-06,
748
+ "loss": 0.6753,
749
+ "step": 98
750
+ },
751
+ {
752
+ "epoch": 49.5,
753
+ "learning_rate": 3.092105263157895e-06,
754
+ "loss": 0.6628,
755
+ "step": 99
756
+ },
757
+ {
758
+ "epoch": 50.0,
759
+ "learning_rate": 3.0701754385964915e-06,
760
+ "loss": 0.6908,
761
+ "step": 100
762
+ },
763
+ {
764
+ "epoch": 50.0,
765
+ "eval_accuracy_dropoff": 0.6332025963898255,
766
+ "eval_accuracy_undropoff": 0.8852680210591884,
767
+ "eval_accuracy_unlabeled": NaN,
768
+ "eval_iou_dropoff": 0.3066633490254799,
769
+ "eval_iou_undropoff": 0.872782516292678,
770
+ "eval_iou_unlabeled": 0.0,
771
+ "eval_loss": 0.6750925779342651,
772
+ "eval_mean_accuracy": 0.759235308724507,
773
+ "eval_mean_iou": 0.3931486217727193,
774
+ "eval_overall_accuracy": 0.8747873306274414,
775
+ "eval_runtime": 2.9141,
776
+ "eval_samples_per_second": 6.863,
777
+ "eval_steps_per_second": 0.686,
778
+ "step": 100
779
+ },
780
+ {
781
+ "epoch": 50.5,
782
+ "learning_rate": 3.048245614035088e-06,
783
+ "loss": 0.674,
784
+ "step": 101
785
+ },
786
+ {
787
+ "epoch": 51.0,
788
+ "learning_rate": 3.0263157894736843e-06,
789
+ "loss": 0.7057,
790
+ "step": 102
791
+ },
792
+ {
793
+ "epoch": 51.5,
794
+ "learning_rate": 3.004385964912281e-06,
795
+ "loss": 0.6581,
796
+ "step": 103
797
+ },
798
+ {
799
+ "epoch": 52.0,
800
+ "learning_rate": 2.9824561403508774e-06,
801
+ "loss": 0.6645,
802
+ "step": 104
803
+ },
804
+ {
805
+ "epoch": 52.5,
806
+ "learning_rate": 2.960526315789474e-06,
807
+ "loss": 0.711,
808
+ "step": 105
809
+ },
810
+ {
811
+ "epoch": 53.0,
812
+ "learning_rate": 2.9385964912280706e-06,
813
+ "loss": 0.6255,
814
+ "step": 106
815
+ },
816
+ {
817
+ "epoch": 53.5,
818
+ "learning_rate": 2.916666666666667e-06,
819
+ "loss": 0.6572,
820
+ "step": 107
821
+ },
822
+ {
823
+ "epoch": 54.0,
824
+ "learning_rate": 2.8947368421052634e-06,
825
+ "loss": 0.6244,
826
+ "step": 108
827
+ },
828
+ {
829
+ "epoch": 54.5,
830
+ "learning_rate": 2.8728070175438597e-06,
831
+ "loss": 0.6323,
832
+ "step": 109
833
+ },
834
+ {
835
+ "epoch": 55.0,
836
+ "learning_rate": 2.8508771929824565e-06,
837
+ "loss": 0.643,
838
+ "step": 110
839
+ },
840
+ {
841
+ "epoch": 55.0,
842
+ "eval_accuracy_dropoff": 0.6194224638179775,
843
+ "eval_accuracy_undropoff": 0.923405212258589,
844
+ "eval_accuracy_unlabeled": NaN,
845
+ "eval_iou_dropoff": 0.3308480506497834,
846
+ "eval_iou_undropoff": 0.9093924657307382,
847
+ "eval_iou_unlabeled": 0.0,
848
+ "eval_loss": 0.6101020574569702,
849
+ "eval_mean_accuracy": 0.7714138380382832,
850
+ "eval_mean_iou": 0.4134135054601738,
851
+ "eval_overall_accuracy": 0.9107658386230468,
852
+ "eval_runtime": 2.8349,
853
+ "eval_samples_per_second": 7.055,
854
+ "eval_steps_per_second": 0.705,
855
+ "step": 110
856
+ },
857
+ {
858
+ "epoch": 55.5,
859
+ "learning_rate": 2.828947368421053e-06,
860
+ "loss": 0.6638,
861
+ "step": 111
862
+ },
863
+ {
864
+ "epoch": 56.0,
865
+ "learning_rate": 2.8070175438596493e-06,
866
+ "loss": 0.5887,
867
+ "step": 112
868
+ },
869
+ {
870
+ "epoch": 56.5,
871
+ "learning_rate": 2.7850877192982456e-06,
872
+ "loss": 0.648,
873
+ "step": 113
874
+ },
875
+ {
876
+ "epoch": 57.0,
877
+ "learning_rate": 2.7631578947368424e-06,
878
+ "loss": 0.578,
879
+ "step": 114
880
+ },
881
+ {
882
+ "epoch": 57.5,
883
+ "learning_rate": 2.741228070175439e-06,
884
+ "loss": 0.6064,
885
+ "step": 115
886
+ },
887
+ {
888
+ "epoch": 58.0,
889
+ "learning_rate": 2.7192982456140356e-06,
890
+ "loss": 0.6784,
891
+ "step": 116
892
+ },
893
+ {
894
+ "epoch": 58.5,
895
+ "learning_rate": 2.697368421052632e-06,
896
+ "loss": 0.624,
897
+ "step": 117
898
+ },
899
+ {
900
+ "epoch": 59.0,
901
+ "learning_rate": 2.6754385964912284e-06,
902
+ "loss": 0.6188,
903
+ "step": 118
904
+ },
905
+ {
906
+ "epoch": 59.5,
907
+ "learning_rate": 2.6535087719298247e-06,
908
+ "loss": 0.607,
909
+ "step": 119
910
+ },
911
+ {
912
+ "epoch": 60.0,
913
+ "learning_rate": 2.631578947368421e-06,
914
+ "loss": 0.6014,
915
+ "step": 120
916
+ },
917
+ {
918
+ "epoch": 60.0,
919
+ "eval_accuracy_dropoff": 0.6338585747379527,
920
+ "eval_accuracy_undropoff": 0.9313126569065759,
921
+ "eval_accuracy_unlabeled": NaN,
922
+ "eval_iou_dropoff": 0.3323511938406625,
923
+ "eval_iou_undropoff": 0.9174802078875154,
924
+ "eval_iou_unlabeled": 0.0,
925
+ "eval_loss": 0.5971364974975586,
926
+ "eval_mean_accuracy": 0.7825856158222644,
927
+ "eval_mean_iou": 0.41661046724272593,
928
+ "eval_overall_accuracy": 0.9189447402954102,
929
+ "eval_runtime": 3.1128,
930
+ "eval_samples_per_second": 6.425,
931
+ "eval_steps_per_second": 0.643,
932
+ "step": 120
933
+ },
934
+ {
935
+ "epoch": 60.5,
936
+ "learning_rate": 2.6096491228070175e-06,
937
+ "loss": 0.5448,
938
+ "step": 121
939
+ },
940
+ {
941
+ "epoch": 61.0,
942
+ "learning_rate": 2.5877192982456147e-06,
943
+ "loss": 0.712,
944
+ "step": 122
945
+ },
946
+ {
947
+ "epoch": 61.5,
948
+ "learning_rate": 2.565789473684211e-06,
949
+ "loss": 0.6238,
950
+ "step": 123
951
+ },
952
+ {
953
+ "epoch": 62.0,
954
+ "learning_rate": 2.5438596491228075e-06,
955
+ "loss": 0.5561,
956
+ "step": 124
957
+ },
958
+ {
959
+ "epoch": 62.5,
960
+ "learning_rate": 2.521929824561404e-06,
961
+ "loss": 0.5693,
962
+ "step": 125
963
+ },
964
+ {
965
+ "epoch": 63.0,
966
+ "learning_rate": 2.5e-06,
967
+ "loss": 0.6064,
968
+ "step": 126
969
+ },
970
+ {
971
+ "epoch": 63.5,
972
+ "learning_rate": 2.4780701754385966e-06,
973
+ "loss": 0.5594,
974
+ "step": 127
975
+ },
976
+ {
977
+ "epoch": 64.0,
978
+ "learning_rate": 2.456140350877193e-06,
979
+ "loss": 0.574,
980
+ "step": 128
981
+ },
982
+ {
983
+ "epoch": 64.5,
984
+ "learning_rate": 2.4342105263157898e-06,
985
+ "loss": 0.5481,
986
+ "step": 129
987
+ },
988
+ {
989
+ "epoch": 65.0,
990
+ "learning_rate": 2.412280701754386e-06,
991
+ "loss": 0.5685,
992
+ "step": 130
993
+ },
994
+ {
995
+ "epoch": 65.0,
996
+ "eval_accuracy_dropoff": 0.6439000894515929,
997
+ "eval_accuracy_undropoff": 0.9453340723220531,
998
+ "eval_accuracy_unlabeled": NaN,
999
+ "eval_iou_dropoff": 0.35985356361855786,
1000
+ "eval_iou_undropoff": 0.9313976351706408,
1001
+ "eval_iou_unlabeled": 0.0,
1002
+ "eval_loss": 0.5594618916511536,
1003
+ "eval_mean_accuracy": 0.794617080886823,
1004
+ "eval_mean_iou": 0.4304170662630662,
1005
+ "eval_overall_accuracy": 0.9328006744384766,
1006
+ "eval_runtime": 3.0082,
1007
+ "eval_samples_per_second": 6.649,
1008
+ "eval_steps_per_second": 0.665,
1009
+ "step": 130
1010
+ },
1011
+ {
1012
+ "epoch": 65.5,
1013
+ "learning_rate": 2.3903508771929825e-06,
1014
+ "loss": 0.5226,
1015
+ "step": 131
1016
+ },
1017
+ {
1018
+ "epoch": 66.0,
1019
+ "learning_rate": 2.368421052631579e-06,
1020
+ "loss": 0.6177,
1021
+ "step": 132
1022
+ },
1023
+ {
1024
+ "epoch": 66.5,
1025
+ "learning_rate": 2.3464912280701757e-06,
1026
+ "loss": 0.5917,
1027
+ "step": 133
1028
+ },
1029
+ {
1030
+ "epoch": 67.0,
1031
+ "learning_rate": 2.324561403508772e-06,
1032
+ "loss": 0.5205,
1033
+ "step": 134
1034
+ },
1035
+ {
1036
+ "epoch": 67.5,
1037
+ "learning_rate": 2.3026315789473684e-06,
1038
+ "loss": 0.5653,
1039
+ "step": 135
1040
+ },
1041
+ {
1042
+ "epoch": 68.0,
1043
+ "learning_rate": 2.280701754385965e-06,
1044
+ "loss": 0.5249,
1045
+ "step": 136
1046
+ },
1047
+ {
1048
+ "epoch": 68.5,
1049
+ "learning_rate": 2.2587719298245616e-06,
1050
+ "loss": 0.4962,
1051
+ "step": 137
1052
+ },
1053
+ {
1054
+ "epoch": 69.0,
1055
+ "learning_rate": 2.236842105263158e-06,
1056
+ "loss": 0.5615,
1057
+ "step": 138
1058
+ },
1059
+ {
1060
+ "epoch": 69.5,
1061
+ "learning_rate": 2.2149122807017543e-06,
1062
+ "loss": 0.4999,
1063
+ "step": 139
1064
+ },
1065
+ {
1066
+ "epoch": 70.0,
1067
+ "learning_rate": 2.192982456140351e-06,
1068
+ "loss": 0.5172,
1069
+ "step": 140
1070
+ },
1071
+ {
1072
+ "epoch": 70.0,
1073
+ "eval_accuracy_dropoff": 0.6487855226037295,
1074
+ "eval_accuracy_undropoff": 0.9532476862654569,
1075
+ "eval_accuracy_unlabeled": NaN,
1076
+ "eval_iou_dropoff": 0.37272109166231837,
1077
+ "eval_iou_undropoff": 0.9393002997549955,
1078
+ "eval_iou_unlabeled": 0.0,
1079
+ "eval_loss": 0.5343549251556396,
1080
+ "eval_mean_accuracy": 0.8010166044345932,
1081
+ "eval_mean_iou": 0.43734046380577124,
1082
+ "eval_overall_accuracy": 0.94058837890625,
1083
+ "eval_runtime": 2.961,
1084
+ "eval_samples_per_second": 6.754,
1085
+ "eval_steps_per_second": 0.675,
1086
+ "step": 140
1087
+ },
1088
+ {
1089
+ "epoch": 70.5,
1090
+ "learning_rate": 2.1710526315789475e-06,
1091
+ "loss": 0.527,
1092
+ "step": 141
1093
+ },
1094
+ {
1095
+ "epoch": 71.0,
1096
+ "learning_rate": 2.149122807017544e-06,
1097
+ "loss": 0.5286,
1098
+ "step": 142
1099
+ },
1100
+ {
1101
+ "epoch": 71.5,
1102
+ "learning_rate": 2.1271929824561407e-06,
1103
+ "loss": 0.5048,
1104
+ "step": 143
1105
+ },
1106
+ {
1107
+ "epoch": 72.0,
1108
+ "learning_rate": 2.105263157894737e-06,
1109
+ "loss": 0.5011,
1110
+ "step": 144
1111
+ },
1112
+ {
1113
+ "epoch": 72.5,
1114
+ "learning_rate": 2.0833333333333334e-06,
1115
+ "loss": 0.512,
1116
+ "step": 145
1117
+ },
1118
+ {
1119
+ "epoch": 73.0,
1120
+ "learning_rate": 2.06140350877193e-06,
1121
+ "loss": 0.5146,
1122
+ "step": 146
1123
+ },
1124
+ {
1125
+ "epoch": 73.5,
1126
+ "learning_rate": 2.0394736842105266e-06,
1127
+ "loss": 0.4928,
1128
+ "step": 147
1129
+ },
1130
+ {
1131
+ "epoch": 74.0,
1132
+ "learning_rate": 2.017543859649123e-06,
1133
+ "loss": 0.5346,
1134
+ "step": 148
1135
+ },
1136
+ {
1137
+ "epoch": 74.5,
1138
+ "learning_rate": 1.9956140350877194e-06,
1139
+ "loss": 0.5102,
1140
+ "step": 149
1141
+ },
1142
+ {
1143
+ "epoch": 75.0,
1144
+ "learning_rate": 1.973684210526316e-06,
1145
+ "loss": 0.4757,
1146
+ "step": 150
1147
+ },
1148
+ {
1149
+ "epoch": 75.0,
1150
+ "eval_accuracy_dropoff": 0.6368265327186403,
1151
+ "eval_accuracy_undropoff": 0.9625887955644756,
1152
+ "eval_accuracy_unlabeled": NaN,
1153
+ "eval_iou_dropoff": 0.38222216103853196,
1154
+ "eval_iou_undropoff": 0.947906194306642,
1155
+ "eval_iou_unlabeled": 0.0,
1156
+ "eval_loss": 0.4962589144706726,
1157
+ "eval_mean_accuracy": 0.7997076641415579,
1158
+ "eval_mean_iou": 0.4433761184483913,
1159
+ "eval_overall_accuracy": 0.9490438461303711,
1160
+ "eval_runtime": 2.9269,
1161
+ "eval_samples_per_second": 6.833,
1162
+ "eval_steps_per_second": 0.683,
1163
+ "step": 150
1164
+ },
1165
+ {
1166
+ "epoch": 75.5,
1167
+ "learning_rate": 1.9517543859649125e-06,
1168
+ "loss": 0.5084,
1169
+ "step": 151
1170
+ },
1171
+ {
1172
+ "epoch": 76.0,
1173
+ "learning_rate": 1.929824561403509e-06,
1174
+ "loss": 0.4696,
1175
+ "step": 152
1176
+ },
1177
+ {
1178
+ "epoch": 76.5,
1179
+ "learning_rate": 1.9078947368421057e-06,
1180
+ "loss": 0.5123,
1181
+ "step": 153
1182
+ },
1183
+ {
1184
+ "epoch": 77.0,
1185
+ "learning_rate": 1.8859649122807019e-06,
1186
+ "loss": 0.4738,
1187
+ "step": 154
1188
+ },
1189
+ {
1190
+ "epoch": 77.5,
1191
+ "learning_rate": 1.8640350877192984e-06,
1192
+ "loss": 0.5085,
1193
+ "step": 155
1194
+ },
1195
+ {
1196
+ "epoch": 78.0,
1197
+ "learning_rate": 1.8421052631578948e-06,
1198
+ "loss": 0.4322,
1199
+ "step": 156
1200
+ },
1201
+ {
1202
+ "epoch": 78.5,
1203
+ "learning_rate": 1.8201754385964914e-06,
1204
+ "loss": 0.4899,
1205
+ "step": 157
1206
+ },
1207
+ {
1208
+ "epoch": 79.0,
1209
+ "learning_rate": 1.798245614035088e-06,
1210
+ "loss": 0.4354,
1211
+ "step": 158
1212
+ },
1213
+ {
1214
+ "epoch": 79.5,
1215
+ "learning_rate": 1.7763157894736844e-06,
1216
+ "loss": 0.5503,
1217
+ "step": 159
1218
+ },
1219
+ {
1220
+ "epoch": 80.0,
1221
+ "learning_rate": 1.7543859649122807e-06,
1222
+ "loss": 0.4288,
1223
+ "step": 160
1224
+ },
1225
+ {
1226
+ "epoch": 80.0,
1227
+ "eval_accuracy_dropoff": 0.6168673593431042,
1228
+ "eval_accuracy_undropoff": 0.970246284243321,
1229
+ "eval_accuracy_unlabeled": NaN,
1230
+ "eval_iou_dropoff": 0.3917943279355294,
1231
+ "eval_iou_undropoff": 0.9545508298763311,
1232
+ "eval_iou_unlabeled": 0.0,
1233
+ "eval_loss": 0.45985984802246094,
1234
+ "eval_mean_accuracy": 0.7935568217932126,
1235
+ "eval_mean_iou": 0.4487817192706201,
1236
+ "eval_overall_accuracy": 0.9555530548095703,
1237
+ "eval_runtime": 3.3564,
1238
+ "eval_samples_per_second": 5.959,
1239
+ "eval_steps_per_second": 0.596,
1240
+ "step": 160
1241
+ },
1242
+ {
1243
+ "epoch": 80.5,
1244
+ "learning_rate": 1.7324561403508775e-06,
1245
+ "loss": 0.4411,
1246
+ "step": 161
1247
+ },
1248
+ {
1249
+ "epoch": 81.0,
1250
+ "learning_rate": 1.710526315789474e-06,
1251
+ "loss": 0.4471,
1252
+ "step": 162
1253
+ },
1254
+ {
1255
+ "epoch": 81.5,
1256
+ "learning_rate": 1.6885964912280703e-06,
1257
+ "loss": 0.4561,
1258
+ "step": 163
1259
+ },
1260
+ {
1261
+ "epoch": 82.0,
1262
+ "learning_rate": 1.6666666666666667e-06,
1263
+ "loss": 0.4375,
1264
+ "step": 164
1265
+ },
1266
+ {
1267
+ "epoch": 82.5,
1268
+ "learning_rate": 1.6447368421052635e-06,
1269
+ "loss": 0.4263,
1270
+ "step": 165
1271
+ },
1272
+ {
1273
+ "epoch": 83.0,
1274
+ "learning_rate": 1.6228070175438598e-06,
1275
+ "loss": 0.4892,
1276
+ "step": 166
1277
+ },
1278
+ {
1279
+ "epoch": 83.5,
1280
+ "learning_rate": 1.6008771929824562e-06,
1281
+ "loss": 0.4613,
1282
+ "step": 167
1283
+ },
1284
+ {
1285
+ "epoch": 84.0,
1286
+ "learning_rate": 1.5789473684210526e-06,
1287
+ "loss": 0.4389,
1288
+ "step": 168
1289
+ },
1290
+ {
1291
+ "epoch": 84.5,
1292
+ "learning_rate": 1.5570175438596494e-06,
1293
+ "loss": 0.5088,
1294
+ "step": 169
1295
+ },
1296
+ {
1297
+ "epoch": 85.0,
1298
+ "learning_rate": 1.5350877192982458e-06,
1299
+ "loss": 0.4124,
1300
+ "step": 170
1301
+ },
1302
+ {
1303
+ "epoch": 85.0,
1304
+ "eval_accuracy_dropoff": 0.6296428817174706,
1305
+ "eval_accuracy_undropoff": 0.96810195656219,
1306
+ "eval_accuracy_unlabeled": NaN,
1307
+ "eval_iou_dropoff": 0.38764311494947556,
1308
+ "eval_iou_undropoff": 0.9529083620698436,
1309
+ "eval_iou_unlabeled": 0.0,
1310
+ "eval_loss": 0.47101444005966187,
1311
+ "eval_mean_accuracy": 0.7988724191398303,
1312
+ "eval_mean_iou": 0.4468504923397731,
1313
+ "eval_overall_accuracy": 0.9540290832519531,
1314
+ "eval_runtime": 3.1781,
1315
+ "eval_samples_per_second": 6.293,
1316
+ "eval_steps_per_second": 0.629,
1317
+ "step": 170
1318
+ },
1319
+ {
1320
+ "epoch": 85.5,
1321
+ "learning_rate": 1.5131578947368421e-06,
1322
+ "loss": 0.444,
1323
+ "step": 171
1324
+ },
1325
+ {
1326
+ "epoch": 86.0,
1327
+ "learning_rate": 1.4912280701754387e-06,
1328
+ "loss": 0.4318,
1329
+ "step": 172
1330
+ },
1331
+ {
1332
+ "epoch": 86.5,
1333
+ "learning_rate": 1.4692982456140353e-06,
1334
+ "loss": 0.441,
1335
+ "step": 173
1336
+ },
1337
+ {
1338
+ "epoch": 87.0,
1339
+ "learning_rate": 1.4473684210526317e-06,
1340
+ "loss": 0.4186,
1341
+ "step": 174
1342
+ },
1343
+ {
1344
+ "epoch": 87.5,
1345
+ "learning_rate": 1.4254385964912283e-06,
1346
+ "loss": 0.3905,
1347
+ "step": 175
1348
+ },
1349
+ {
1350
+ "epoch": 88.0,
1351
+ "learning_rate": 1.4035087719298246e-06,
1352
+ "loss": 0.4534,
1353
+ "step": 176
1354
+ },
1355
+ {
1356
+ "epoch": 88.5,
1357
+ "learning_rate": 1.3815789473684212e-06,
1358
+ "loss": 0.4197,
1359
+ "step": 177
1360
+ },
1361
+ {
1362
+ "epoch": 89.0,
1363
+ "learning_rate": 1.3596491228070178e-06,
1364
+ "loss": 0.421,
1365
+ "step": 178
1366
+ },
1367
+ {
1368
+ "epoch": 89.5,
1369
+ "learning_rate": 1.3377192982456142e-06,
1370
+ "loss": 0.4123,
1371
+ "step": 179
1372
+ },
1373
+ {
1374
+ "epoch": 90.0,
1375
+ "learning_rate": 1.3157894736842106e-06,
1376
+ "loss": 0.4995,
1377
+ "step": 180
1378
+ },
1379
+ {
1380
+ "epoch": 90.0,
1381
+ "eval_accuracy_dropoff": 0.6004220280281658,
1382
+ "eval_accuracy_undropoff": 0.9762418443407163,
1383
+ "eval_accuracy_unlabeled": NaN,
1384
+ "eval_iou_dropoff": 0.4015369512531828,
1385
+ "eval_iou_undropoff": 0.9596764001720002,
1386
+ "eval_iou_unlabeled": 0.0,
1387
+ "eval_loss": 0.42094817757606506,
1388
+ "eval_mean_accuracy": 0.788331936184441,
1389
+ "eval_mean_iou": 0.4537377838083943,
1390
+ "eval_overall_accuracy": 0.9606155395507813,
1391
+ "eval_runtime": 3.0265,
1392
+ "eval_samples_per_second": 6.608,
1393
+ "eval_steps_per_second": 0.661,
1394
+ "step": 180
1395
+ },
1396
+ {
1397
+ "epoch": 90.5,
1398
+ "learning_rate": 1.2938596491228074e-06,
1399
+ "loss": 0.3708,
1400
+ "step": 181
1401
+ },
1402
+ {
1403
+ "epoch": 91.0,
1404
+ "learning_rate": 1.2719298245614037e-06,
1405
+ "loss": 0.5003,
1406
+ "step": 182
1407
+ },
1408
+ {
1409
+ "epoch": 91.5,
1410
+ "learning_rate": 1.25e-06,
1411
+ "loss": 0.3846,
1412
+ "step": 183
1413
+ },
1414
+ {
1415
+ "epoch": 92.0,
1416
+ "learning_rate": 1.2280701754385965e-06,
1417
+ "loss": 0.4393,
1418
+ "step": 184
1419
+ },
1420
+ {
1421
+ "epoch": 92.5,
1422
+ "learning_rate": 1.206140350877193e-06,
1423
+ "loss": 0.416,
1424
+ "step": 185
1425
+ },
1426
+ {
1427
+ "epoch": 93.0,
1428
+ "learning_rate": 1.1842105263157894e-06,
1429
+ "loss": 0.3925,
1430
+ "step": 186
1431
+ },
1432
+ {
1433
+ "epoch": 93.5,
1434
+ "learning_rate": 1.162280701754386e-06,
1435
+ "loss": 0.4545,
1436
+ "step": 187
1437
+ },
1438
+ {
1439
+ "epoch": 94.0,
1440
+ "learning_rate": 1.1403508771929824e-06,
1441
+ "loss": 0.3882,
1442
+ "step": 188
1443
+ },
1444
+ {
1445
+ "epoch": 94.5,
1446
+ "learning_rate": 1.118421052631579e-06,
1447
+ "loss": 0.4603,
1448
+ "step": 189
1449
+ },
1450
+ {
1451
+ "epoch": 95.0,
1452
+ "learning_rate": 1.0964912280701756e-06,
1453
+ "loss": 0.3815,
1454
+ "step": 190
1455
+ },
1456
+ {
1457
+ "epoch": 95.0,
1458
+ "eval_accuracy_dropoff": 0.6089635083373471,
1459
+ "eval_accuracy_undropoff": 0.9747594223549395,
1460
+ "eval_accuracy_unlabeled": NaN,
1461
+ "eval_iou_dropoff": 0.39879296567552464,
1462
+ "eval_iou_undropoff": 0.9585519879558038,
1463
+ "eval_iou_unlabeled": 0.0,
1464
+ "eval_loss": 0.4287094175815582,
1465
+ "eval_mean_accuracy": 0.7918614653461433,
1466
+ "eval_mean_iou": 0.4524483178771095,
1467
+ "eval_overall_accuracy": 0.9595499038696289,
1468
+ "eval_runtime": 2.9721,
1469
+ "eval_samples_per_second": 6.729,
1470
+ "eval_steps_per_second": 0.673,
1471
+ "step": 190
1472
+ },
1473
+ {
1474
+ "epoch": 95.5,
1475
+ "learning_rate": 1.074561403508772e-06,
1476
+ "loss": 0.4023,
1477
+ "step": 191
1478
+ },
1479
+ {
1480
+ "epoch": 96.0,
1481
+ "learning_rate": 1.0526315789473685e-06,
1482
+ "loss": 0.4105,
1483
+ "step": 192
1484
+ },
1485
+ {
1486
+ "epoch": 96.5,
1487
+ "learning_rate": 1.030701754385965e-06,
1488
+ "loss": 0.4662,
1489
+ "step": 193
1490
+ },
1491
+ {
1492
+ "epoch": 97.0,
1493
+ "learning_rate": 1.0087719298245615e-06,
1494
+ "loss": 0.392,
1495
+ "step": 194
1496
+ },
1497
+ {
1498
+ "epoch": 97.5,
1499
+ "learning_rate": 9.86842105263158e-07,
1500
+ "loss": 0.3692,
1501
+ "step": 195
1502
+ },
1503
+ {
1504
+ "epoch": 98.0,
1505
+ "learning_rate": 9.649122807017545e-07,
1506
+ "loss": 0.4193,
1507
+ "step": 196
1508
+ },
1509
+ {
1510
+ "epoch": 98.5,
1511
+ "learning_rate": 9.429824561403509e-07,
1512
+ "loss": 0.3569,
1513
+ "step": 197
1514
+ },
1515
+ {
1516
+ "epoch": 99.0,
1517
+ "learning_rate": 9.210526315789474e-07,
1518
+ "loss": 0.4784,
1519
+ "step": 198
1520
+ },
1521
+ {
1522
+ "epoch": 99.5,
1523
+ "learning_rate": 8.99122807017544e-07,
1524
+ "loss": 0.4157,
1525
+ "step": 199
1526
+ },
1527
+ {
1528
+ "epoch": 100.0,
1529
+ "learning_rate": 8.771929824561404e-07,
1530
+ "loss": 0.3764,
1531
+ "step": 200
1532
+ },
1533
+ {
1534
+ "epoch": 100.0,
1535
+ "eval_accuracy_dropoff": 0.6073120943140897,
1536
+ "eval_accuracy_undropoff": 0.9753421222575243,
1537
+ "eval_accuracy_unlabeled": NaN,
1538
+ "eval_iou_dropoff": 0.39976024832717344,
1539
+ "eval_iou_undropoff": 0.9590456081576902,
1540
+ "eval_iou_unlabeled": 0.0,
1541
+ "eval_loss": 0.4244542717933655,
1542
+ "eval_mean_accuracy": 0.791327108285807,
1543
+ "eval_mean_iou": 0.4529352854949546,
1544
+ "eval_overall_accuracy": 0.9600397109985351,
1545
+ "eval_runtime": 2.942,
1546
+ "eval_samples_per_second": 6.798,
1547
+ "eval_steps_per_second": 0.68,
1548
+ "step": 200
1549
+ },
1550
+ {
1551
+ "epoch": 100.5,
1552
+ "learning_rate": 8.55263157894737e-07,
1553
+ "loss": 0.3986,
1554
+ "step": 201
1555
+ },
1556
+ {
1557
+ "epoch": 101.0,
1558
+ "learning_rate": 8.333333333333333e-07,
1559
+ "loss": 0.3613,
1560
+ "step": 202
1561
+ },
1562
+ {
1563
+ "epoch": 101.5,
1564
+ "learning_rate": 8.114035087719299e-07,
1565
+ "loss": 0.408,
1566
+ "step": 203
1567
+ },
1568
+ {
1569
+ "epoch": 102.0,
1570
+ "learning_rate": 7.894736842105263e-07,
1571
+ "loss": 0.3586,
1572
+ "step": 204
1573
+ },
1574
+ {
1575
+ "epoch": 102.5,
1576
+ "learning_rate": 7.675438596491229e-07,
1577
+ "loss": 0.3742,
1578
+ "step": 205
1579
+ },
1580
+ {
1581
+ "epoch": 103.0,
1582
+ "learning_rate": 7.456140350877194e-07,
1583
+ "loss": 0.4415,
1584
+ "step": 206
1585
+ },
1586
+ {
1587
+ "epoch": 103.5,
1588
+ "learning_rate": 7.236842105263158e-07,
1589
+ "loss": 0.4422,
1590
+ "step": 207
1591
+ },
1592
+ {
1593
+ "epoch": 104.0,
1594
+ "learning_rate": 7.017543859649123e-07,
1595
+ "loss": 0.3713,
1596
+ "step": 208
1597
+ },
1598
+ {
1599
+ "epoch": 104.5,
1600
+ "learning_rate": 6.798245614035089e-07,
1601
+ "loss": 0.3961,
1602
+ "step": 209
1603
+ },
1604
+ {
1605
+ "epoch": 105.0,
1606
+ "learning_rate": 6.578947368421053e-07,
1607
+ "loss": 0.4074,
1608
+ "step": 210
1609
+ },
1610
+ {
1611
+ "epoch": 105.0,
1612
+ "eval_accuracy_dropoff": 0.6018211426867589,
1613
+ "eval_accuracy_undropoff": 0.9768896203594709,
1614
+ "eval_accuracy_unlabeled": NaN,
1615
+ "eval_iou_dropoff": 0.4021185752336348,
1616
+ "eval_iou_undropoff": 0.9603319715732418,
1617
+ "eval_iou_unlabeled": 0.0,
1618
+ "eval_loss": 0.4095707833766937,
1619
+ "eval_mean_accuracy": 0.7893553815231149,
1620
+ "eval_mean_iou": 0.45415018226895887,
1621
+ "eval_overall_accuracy": 0.9612945556640625,
1622
+ "eval_runtime": 2.9806,
1623
+ "eval_samples_per_second": 6.71,
1624
+ "eval_steps_per_second": 0.671,
1625
+ "step": 210
1626
+ },
1627
+ {
1628
+ "epoch": 105.5,
1629
+ "learning_rate": 6.359649122807019e-07,
1630
+ "loss": 0.399,
1631
+ "step": 211
1632
+ },
1633
+ {
1634
+ "epoch": 106.0,
1635
+ "learning_rate": 6.140350877192982e-07,
1636
+ "loss": 0.3553,
1637
+ "step": 212
1638
+ },
1639
+ {
1640
+ "epoch": 106.5,
1641
+ "learning_rate": 5.921052631578947e-07,
1642
+ "loss": 0.3877,
1643
+ "step": 213
1644
+ },
1645
+ {
1646
+ "epoch": 107.0,
1647
+ "learning_rate": 5.701754385964912e-07,
1648
+ "loss": 0.3452,
1649
+ "step": 214
1650
+ },
1651
+ {
1652
+ "epoch": 107.5,
1653
+ "learning_rate": 5.482456140350878e-07,
1654
+ "loss": 0.3324,
1655
+ "step": 215
1656
+ },
1657
+ {
1658
+ "epoch": 108.0,
1659
+ "learning_rate": 5.263157894736843e-07,
1660
+ "loss": 0.4387,
1661
+ "step": 216
1662
+ },
1663
+ {
1664
+ "epoch": 108.5,
1665
+ "learning_rate": 5.043859649122807e-07,
1666
+ "loss": 0.4437,
1667
+ "step": 217
1668
+ },
1669
+ {
1670
+ "epoch": 109.0,
1671
+ "learning_rate": 4.824561403508772e-07,
1672
+ "loss": 0.364,
1673
+ "step": 218
1674
+ },
1675
+ {
1676
+ "epoch": 109.5,
1677
+ "learning_rate": 4.605263157894737e-07,
1678
+ "loss": 0.4003,
1679
+ "step": 219
1680
+ },
1681
+ {
1682
+ "epoch": 110.0,
1683
+ "learning_rate": 4.385964912280702e-07,
1684
+ "loss": 0.3975,
1685
+ "step": 220
1686
+ },
1687
+ {
1688
+ "epoch": 110.0,
1689
+ "eval_accuracy_dropoff": 0.6044679923851465,
1690
+ "eval_accuracy_undropoff": 0.9765160794724655,
1691
+ "eval_accuracy_unlabeled": NaN,
1692
+ "eval_iou_dropoff": 0.4012710728902748,
1693
+ "eval_iou_undropoff": 0.9600690045543241,
1694
+ "eval_iou_unlabeled": 0.0,
1695
+ "eval_loss": 0.4107142388820648,
1696
+ "eval_mean_accuracy": 0.790492035928806,
1697
+ "eval_mean_iou": 0.4537800258148663,
1698
+ "eval_overall_accuracy": 0.9610466003417969,
1699
+ "eval_runtime": 3.0507,
1700
+ "eval_samples_per_second": 6.556,
1701
+ "eval_steps_per_second": 0.656,
1702
+ "step": 220
1703
+ },
1704
+ {
1705
+ "epoch": 110.5,
1706
+ "learning_rate": 4.1666666666666667e-07,
1707
+ "loss": 0.3876,
1708
+ "step": 221
1709
+ },
1710
+ {
1711
+ "epoch": 111.0,
1712
+ "learning_rate": 3.9473684210526315e-07,
1713
+ "loss": 0.3976,
1714
+ "step": 222
1715
+ },
1716
+ {
1717
+ "epoch": 111.5,
1718
+ "learning_rate": 3.728070175438597e-07,
1719
+ "loss": 0.3826,
1720
+ "step": 223
1721
+ },
1722
+ {
1723
+ "epoch": 112.0,
1724
+ "learning_rate": 3.5087719298245616e-07,
1725
+ "loss": 0.3957,
1726
+ "step": 224
1727
+ },
1728
+ {
1729
+ "epoch": 112.5,
1730
+ "learning_rate": 3.2894736842105264e-07,
1731
+ "loss": 0.3323,
1732
+ "step": 225
1733
+ },
1734
+ {
1735
+ "epoch": 113.0,
1736
+ "learning_rate": 3.070175438596491e-07,
1737
+ "loss": 0.4554,
1738
+ "step": 226
1739
+ },
1740
+ {
1741
+ "epoch": 113.5,
1742
+ "learning_rate": 2.850877192982456e-07,
1743
+ "loss": 0.3326,
1744
+ "step": 227
1745
+ },
1746
+ {
1747
+ "epoch": 114.0,
1748
+ "learning_rate": 2.6315789473684213e-07,
1749
+ "loss": 0.4263,
1750
+ "step": 228
1751
+ },
1752
+ {
1753
+ "epoch": 114.5,
1754
+ "learning_rate": 2.412280701754386e-07,
1755
+ "loss": 0.3712,
1756
+ "step": 229
1757
+ },
1758
+ {
1759
+ "epoch": 115.0,
1760
+ "learning_rate": 2.192982456140351e-07,
1761
+ "loss": 0.3598,
1762
+ "step": 230
1763
+ },
1764
+ {
1765
+ "epoch": 115.0,
1766
+ "eval_accuracy_dropoff": 0.5938760063304204,
1767
+ "eval_accuracy_undropoff": 0.9787475733275488,
1768
+ "eval_accuracy_unlabeled": NaN,
1769
+ "eval_iou_dropoff": 0.40565006094370304,
1770
+ "eval_iou_undropoff": 0.9618219900612324,
1771
+ "eval_iou_unlabeled": 0.0,
1772
+ "eval_loss": 0.39181163907051086,
1773
+ "eval_mean_accuracy": 0.7863117898289846,
1774
+ "eval_mean_iou": 0.45582401700164515,
1775
+ "eval_overall_accuracy": 0.9627449035644531,
1776
+ "eval_runtime": 2.9212,
1777
+ "eval_samples_per_second": 6.847,
1778
+ "eval_steps_per_second": 0.685,
1779
+ "step": 230
1780
+ },
1781
+ {
1782
+ "epoch": 115.5,
1783
+ "learning_rate": 1.9736842105263157e-07,
1784
+ "loss": 0.3428,
1785
+ "step": 231
1786
+ },
1787
+ {
1788
+ "epoch": 116.0,
1789
+ "learning_rate": 1.7543859649122808e-07,
1790
+ "loss": 0.392,
1791
+ "step": 232
1792
+ },
1793
+ {
1794
+ "epoch": 116.5,
1795
+ "learning_rate": 1.5350877192982456e-07,
1796
+ "loss": 0.4413,
1797
+ "step": 233
1798
+ },
1799
+ {
1800
+ "epoch": 117.0,
1801
+ "learning_rate": 1.3157894736842107e-07,
1802
+ "loss": 0.3248,
1803
+ "step": 234
1804
+ },
1805
+ {
1806
+ "epoch": 117.5,
1807
+ "learning_rate": 1.0964912280701755e-07,
1808
+ "loss": 0.3836,
1809
+ "step": 235
1810
+ },
1811
+ {
1812
+ "epoch": 118.0,
1813
+ "learning_rate": 8.771929824561404e-08,
1814
+ "loss": 0.3486,
1815
+ "step": 236
1816
+ },
1817
+ {
1818
+ "epoch": 118.5,
1819
+ "learning_rate": 6.578947368421053e-08,
1820
+ "loss": 0.4099,
1821
+ "step": 237
1822
+ },
1823
+ {
1824
+ "epoch": 119.0,
1825
+ "learning_rate": 4.385964912280702e-08,
1826
+ "loss": 0.3073,
1827
+ "step": 238
1828
+ },
1829
+ {
1830
+ "epoch": 119.5,
1831
+ "learning_rate": 2.192982456140351e-08,
1832
+ "loss": 0.3674,
1833
+ "step": 239
1834
+ },
1835
+ {
1836
+ "epoch": 120.0,
1837
+ "learning_rate": 0.0,
1838
+ "loss": 0.3709,
1839
+ "step": 240
1840
+ },
1841
+ {
1842
+ "epoch": 120.0,
1843
+ "eval_accuracy_dropoff": 0.583917062317943,
1844
+ "eval_accuracy_undropoff": 0.9804954740257737,
1845
+ "eval_accuracy_unlabeled": NaN,
1846
+ "eval_iou_dropoff": 0.40859949282573105,
1847
+ "eval_iou_undropoff": 0.9631331874229299,
1848
+ "eval_iou_unlabeled": 0.0,
1849
+ "eval_loss": 0.3769916594028473,
1850
+ "eval_mean_accuracy": 0.7822062681718585,
1851
+ "eval_mean_iou": 0.4572442267495536,
1852
+ "eval_overall_accuracy": 0.9640060424804687,
1853
+ "eval_runtime": 2.829,
1854
+ "eval_samples_per_second": 7.07,
1855
+ "eval_steps_per_second": 0.707,
1856
+ "step": 240
1857
+ }
1858
+ ],
1859
+ "max_steps": 240,
1860
+ "num_train_epochs": 120,
1861
+ "total_flos": 1.3891198199110042e+18,
1862
+ "trial_name": null,
1863
+ "trial_params": null
1864
+ }
{checkpoint-40 β†’ checkpoint-240}/training_args.bin RENAMED
File without changes
checkpoint-40/trainer_state.json DELETED
@@ -1,324 +0,0 @@
1
- {
2
- "best_metric": 0.9755474328994751,
3
- "best_model_checkpoint": "/robodata/smodak/Projects/nspl/scripts/terrainseg/training/models/dropoff-utcustom-train-SF-RGB-b5_3/checkpoint-40",
4
- "epoch": 20.0,
5
- "global_step": 40,
6
- "is_hyper_param_search": false,
7
- "is_local_process_zero": true,
8
- "is_world_process_zero": true,
9
- "log_history": [
10
- {
11
- "epoch": 0.5,
12
- "learning_rate": 4.1666666666666667e-07,
13
- "loss": 1.3484,
14
- "step": 1
15
- },
16
- {
17
- "epoch": 1.0,
18
- "learning_rate": 8.333333333333333e-07,
19
- "loss": 1.3481,
20
- "step": 2
21
- },
22
- {
23
- "epoch": 1.5,
24
- "learning_rate": 1.25e-06,
25
- "loss": 1.3531,
26
- "step": 3
27
- },
28
- {
29
- "epoch": 2.0,
30
- "learning_rate": 1.6666666666666667e-06,
31
- "loss": 1.3399,
32
- "step": 4
33
- },
34
- {
35
- "epoch": 2.5,
36
- "learning_rate": 2.0833333333333334e-06,
37
- "loss": 1.3398,
38
- "step": 5
39
- },
40
- {
41
- "epoch": 3.0,
42
- "learning_rate": 2.5e-06,
43
- "loss": 1.3386,
44
- "step": 6
45
- },
46
- {
47
- "epoch": 3.5,
48
- "learning_rate": 2.916666666666667e-06,
49
- "loss": 1.3376,
50
- "step": 7
51
- },
52
- {
53
- "epoch": 4.0,
54
- "learning_rate": 3.3333333333333333e-06,
55
- "loss": 1.3249,
56
- "step": 8
57
- },
58
- {
59
- "epoch": 4.5,
60
- "learning_rate": 3.7500000000000005e-06,
61
- "loss": 1.3205,
62
- "step": 9
63
- },
64
- {
65
- "epoch": 5.0,
66
- "learning_rate": 4.166666666666667e-06,
67
- "loss": 1.3135,
68
- "step": 10
69
- },
70
- {
71
- "epoch": 5.0,
72
- "eval_accuracy_dropoff": 0.4068717172412211,
73
- "eval_accuracy_undropoff": 0.11034322178517518,
74
- "eval_accuracy_unlabeled": NaN,
75
- "eval_iou_dropoff": 0.053473330857492875,
76
- "eval_iou_undropoff": 0.1102453041626867,
77
- "eval_iou_unlabeled": 0.0,
78
- "eval_loss": 1.2007640600204468,
79
- "eval_mean_accuracy": 0.25860746951319813,
80
- "eval_mean_iou": 0.05457287834005986,
81
- "eval_overall_accuracy": 0.12267265319824219,
82
- "eval_runtime": 2.5511,
83
- "eval_samples_per_second": 7.84,
84
- "eval_steps_per_second": 0.784,
85
- "step": 10
86
- },
87
- {
88
- "epoch": 5.5,
89
- "learning_rate": 4.583333333333333e-06,
90
- "loss": 1.3089,
91
- "step": 11
92
- },
93
- {
94
- "epoch": 6.0,
95
- "learning_rate": 5e-06,
96
- "loss": 1.3043,
97
- "step": 12
98
- },
99
- {
100
- "epoch": 6.5,
101
- "learning_rate": 4.978070175438597e-06,
102
- "loss": 1.2871,
103
- "step": 13
104
- },
105
- {
106
- "epoch": 7.0,
107
- "learning_rate": 4.956140350877193e-06,
108
- "loss": 1.2832,
109
- "step": 14
110
- },
111
- {
112
- "epoch": 7.5,
113
- "learning_rate": 4.9342105263157895e-06,
114
- "loss": 1.2709,
115
- "step": 15
116
- },
117
- {
118
- "epoch": 8.0,
119
- "learning_rate": 4.912280701754386e-06,
120
- "loss": 1.2634,
121
- "step": 16
122
- },
123
- {
124
- "epoch": 8.5,
125
- "learning_rate": 4.890350877192983e-06,
126
- "loss": 1.257,
127
- "step": 17
128
- },
129
- {
130
- "epoch": 9.0,
131
- "learning_rate": 4.8684210526315795e-06,
132
- "loss": 1.2403,
133
- "step": 18
134
- },
135
- {
136
- "epoch": 9.5,
137
- "learning_rate": 4.846491228070176e-06,
138
- "loss": 1.2398,
139
- "step": 19
140
- },
141
- {
142
- "epoch": 10.0,
143
- "learning_rate": 4.824561403508772e-06,
144
- "loss": 1.2309,
145
- "step": 20
146
- },
147
- {
148
- "epoch": 10.0,
149
- "eval_accuracy_dropoff": 0.43877153145714354,
150
- "eval_accuracy_undropoff": 0.24072332003617994,
151
- "eval_accuracy_unlabeled": NaN,
152
- "eval_iou_dropoff": 0.11288106949650205,
153
- "eval_iou_undropoff": 0.24004849364289937,
154
- "eval_iou_unlabeled": 0.0,
155
- "eval_loss": 1.1293678283691406,
156
- "eval_mean_accuracy": 0.3397474257466617,
157
- "eval_mean_iou": 0.11764318771313381,
158
- "eval_overall_accuracy": 0.24895801544189453,
159
- "eval_runtime": 2.6298,
160
- "eval_samples_per_second": 7.605,
161
- "eval_steps_per_second": 0.761,
162
- "step": 20
163
- },
164
- {
165
- "epoch": 10.5,
166
- "learning_rate": 4.802631578947369e-06,
167
- "loss": 1.2159,
168
- "step": 21
169
- },
170
- {
171
- "epoch": 11.0,
172
- "learning_rate": 4.780701754385965e-06,
173
- "loss": 1.2243,
174
- "step": 22
175
- },
176
- {
177
- "epoch": 11.5,
178
- "learning_rate": 4.758771929824561e-06,
179
- "loss": 1.2094,
180
- "step": 23
181
- },
182
- {
183
- "epoch": 12.0,
184
- "learning_rate": 4.736842105263158e-06,
185
- "loss": 1.1904,
186
- "step": 24
187
- },
188
- {
189
- "epoch": 12.5,
190
- "learning_rate": 4.714912280701755e-06,
191
- "loss": 1.1865,
192
- "step": 25
193
- },
194
- {
195
- "epoch": 13.0,
196
- "learning_rate": 4.692982456140351e-06,
197
- "loss": 1.1715,
198
- "step": 26
199
- },
200
- {
201
- "epoch": 13.5,
202
- "learning_rate": 4.671052631578948e-06,
203
- "loss": 1.1722,
204
- "step": 27
205
- },
206
- {
207
- "epoch": 14.0,
208
- "learning_rate": 4.649122807017544e-06,
209
- "loss": 1.1511,
210
- "step": 28
211
- },
212
- {
213
- "epoch": 14.5,
214
- "learning_rate": 4.6271929824561405e-06,
215
- "loss": 1.1512,
216
- "step": 29
217
- },
218
- {
219
- "epoch": 15.0,
220
- "learning_rate": 4.605263157894737e-06,
221
- "loss": 1.1346,
222
- "step": 30
223
- },
224
- {
225
- "epoch": 15.0,
226
- "eval_accuracy_dropoff": 0.46940067432739285,
227
- "eval_accuracy_undropoff": 0.5035900324087019,
228
- "eval_accuracy_unlabeled": NaN,
229
- "eval_iou_dropoff": 0.15243042262584855,
230
- "eval_iou_undropoff": 0.4989255427443525,
231
- "eval_iou_unlabeled": 0.0,
232
- "eval_loss": 1.039512276649475,
233
- "eval_mean_accuracy": 0.4864953533680474,
234
- "eval_mean_iou": 0.21711865512340034,
235
- "eval_overall_accuracy": 0.5021684646606446,
236
- "eval_runtime": 2.9463,
237
- "eval_samples_per_second": 6.788,
238
- "eval_steps_per_second": 0.679,
239
- "step": 30
240
- },
241
- {
242
- "epoch": 15.5,
243
- "learning_rate": 4.583333333333333e-06,
244
- "loss": 1.1343,
245
- "step": 31
246
- },
247
- {
248
- "epoch": 16.0,
249
- "learning_rate": 4.56140350877193e-06,
250
- "loss": 1.1182,
251
- "step": 32
252
- },
253
- {
254
- "epoch": 16.5,
255
- "learning_rate": 4.539473684210527e-06,
256
- "loss": 1.132,
257
- "step": 33
258
- },
259
- {
260
- "epoch": 17.0,
261
- "learning_rate": 4.517543859649123e-06,
262
- "loss": 1.0986,
263
- "step": 34
264
- },
265
- {
266
- "epoch": 17.5,
267
- "learning_rate": 4.4956140350877196e-06,
268
- "loss": 1.0887,
269
- "step": 35
270
- },
271
- {
272
- "epoch": 18.0,
273
- "learning_rate": 4.473684210526316e-06,
274
- "loss": 1.1214,
275
- "step": 36
276
- },
277
- {
278
- "epoch": 18.5,
279
- "learning_rate": 4.451754385964912e-06,
280
- "loss": 1.0945,
281
- "step": 37
282
- },
283
- {
284
- "epoch": 19.0,
285
- "learning_rate": 4.429824561403509e-06,
286
- "loss": 1.076,
287
- "step": 38
288
- },
289
- {
290
- "epoch": 19.5,
291
- "learning_rate": 4.407894736842105e-06,
292
- "loss": 1.0444,
293
- "step": 39
294
- },
295
- {
296
- "epoch": 20.0,
297
- "learning_rate": 4.385964912280702e-06,
298
- "loss": 1.1088,
299
- "step": 40
300
- },
301
- {
302
- "epoch": 20.0,
303
- "eval_accuracy_dropoff": 0.48076332025963897,
304
- "eval_accuracy_undropoff": 0.6235115828521449,
305
- "eval_accuracy_unlabeled": NaN,
306
- "eval_iou_dropoff": 0.1660792839847365,
307
- "eval_iou_undropoff": 0.616259059512146,
308
- "eval_iou_unlabeled": 0.0,
309
- "eval_loss": 0.9755474328994751,
310
- "eval_mean_accuracy": 0.5521374515558919,
311
- "eval_mean_iou": 0.2607794478322942,
312
- "eval_overall_accuracy": 0.6175762176513672,
313
- "eval_runtime": 3.0937,
314
- "eval_samples_per_second": 6.465,
315
- "eval_steps_per_second": 0.646,
316
- "step": 40
317
- }
318
- ],
319
- "max_steps": 240,
320
- "num_train_epochs": 120,
321
- "total_flos": 2.3151996998516736e+17,
322
- "trial_name": null,
323
- "trial_params": null
324
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:884c6c90a3f870941ff5e380a2d4f6d09ca4c8fa633958768d32a3e323aff19c
3
  size 338796489
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7123f1ef46597d6e8e27beb20eeb550ccffbb057b64c28f5346b0a7cf308bc60
3
  size 338796489
runs/Feb12_08-24-49_robovision.csres.utexas.edu/events.out.tfevents.1707747895.robovision.csres.utexas.edu.2833461.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:31910d97c93e8bd05cbd2d4661b405fcdc9b00464fffb122d7d858a144f9477e
3
- size 51487
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8a7d57ce672bb92159df7d0511b3caa6a0c13a39b03e0abfa8975d8780954be
3
+ size 60939